repo
stringclasses
1 value
github_id
int64
3.77B
4.38B
github_node_id
stringlengths
18
24
number
int64
75
77.8k
html_url
stringlengths
46
49
api_url
stringlengths
56
59
title
stringlengths
1
746
body
stringlengths
1
55.7k
state
stringclasses
2 values
state_reason
stringclasses
4 values
locked
bool
2 classes
comments_count
int64
0
128
labels
listlengths
0
5
assignees
listlengths
0
2
created_at
stringdate
2026-01-01 16:55:34
2026-05-05 11:59:03
updated_at
stringdate
2026-03-24 00:13:29
2026-05-05 11:59:15
closed_at
stringlengths
20
20
author_association
stringclasses
3 values
milestone_title
stringclasses
0 values
snapshot_id
stringclasses
76 values
extracted_at
stringdate
2026-03-24 23:36:49
2026-05-05 12:00:33
author_login
stringlengths
3
38
author_id
int64
346
282M
author_node_id
stringlengths
12
20
author_type
stringclasses
2 values
author_site_admin
bool
2 classes
openclaw/openclaw
4,239,066,852
I_kwDOQb6kR878qwbk
64,292
https://github.com/openclaw/openclaw/issues/64292
https://api.github.com/repos/openclaw/openclaw/issues/64292
[Bug]: [2026.4.9] Spawn failed: agentId is not allowed for sessions_spawn (allowed: none)
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary Running /subagents spawn fails with "agentId is not allowed for sessions_spawn (allowed: none)" on OpenClaw 2026.4.9. ### Steps to reproduce 1. Use OpenClaw 2026.4.9 on macOS. 2. Configure a default agent with tools.allow:...
closed
not_planned
true
1
[ "bug", "regression" ]
[]
2026-04-10T12:13:26Z
2026-04-28T05:01:30Z
2026-04-25T16:03:02Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
JiangWanZhengChouYv
246,920,710
U_kgDODre2Bg
User
false
openclaw/openclaw
4,239,074,107
I_kwDOQb6kR878qyM7
64,293
https://github.com/openclaw/openclaw/issues/64293
https://api.github.com/repos/openclaw/openclaw/issues/64293
[Bug]: Heartbeat runs despite heartbeat: {} config (2M+ tokens/day with zero user activity)
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary Heartbeats continue running every 30 minutes despite setting `agents.defaults.heartbeat: {}` in config. This causes constant token burn (~2M input tokens/day) with zero user activity. Config changes and full up...
open
null
false
2
[ "bug", "bug:behavior" ]
[]
2026-04-10T12:14:52Z
2026-05-03T06:49:36Z
null
NONE
null
20260503T090142Z
2026-05-03T09:01:42Z
lucianriediger
46,679,956
MDQ6VXNlcjQ2Njc5OTU2
User
false
openclaw/openclaw
4,239,090,636
I_kwDOQb6kR878q2PM
64,295
https://github.com/openclaw/openclaw/issues/64295
https://api.github.com/repos/openclaw/openclaw/issues/64295
memory-core: dreaming narrative generation fails — idempotencyKey required
## Bug Description The memory-core dreaming system successfully promotes memories to MEMORY.md but fails to generate dream diary entries (DREAMS.md). Every narrative generation attempt for light, REM, and deep phases fails with the same error. ## Error ``` memory-core: narrative generation failed for <phase> phase: ...
closed
completed
true
3
[]
[]
2026-04-10T12:17:41Z
2026-04-15T04:18:56Z
2026-04-12T16:46:23Z
NONE
null
20260417T163222Z
2026-04-17T16:32:22Z
kaishi00-student
217,337,130
U_kgDODPRNKg
User
false
openclaw/openclaw
4,239,098,607
I_kwDOQb6kR878q4Lv
64,296
https://github.com/openclaw/openclaw/issues/64296
https://api.github.com/repos/openclaw/openclaw/issues/64296
[Bug]: WhatsApp Web connection ended before fully opening
### Bug type Crash (process/app exits or hangs) ### Beta release blocker No ### Summary OpenClaw 2026.4.9 issue ### Steps to reproduce run openclaw channels login --channel whatsapp ### Expected behavior OpenClaw 2026.4.9 error openclaw channels login --channel whatsapp ### Actual behavior no issue ### Op...
closed
not_planned
true
3
[ "bug", "bug:crash" ]
[]
2026-04-10T12:19:15Z
2026-04-29T04:32:31Z
2026-04-27T02:26:05Z
NONE
null
20260429T060025Z
2026-04-29T06:00:25Z
athuljayaram
141,337,513
U_kgDOCGyjqQ
User
false
openclaw/openclaw
4,239,147,298
I_kwDOQb6kR878rEEi
64,299
https://github.com/openclaw/openclaw/issues/64299
https://api.github.com/repos/openclaw/openclaw/issues/64299
Bug: feishu_doc write and append actions return 400 on valid markdown content
## Bug Description The `feishu_doc` tool's `write` (replace all) and `append` actions consistently return `400 Bad Request` for any markdown content, even simple plain text. The `update_block` action works correctly with the same content, but `write` and `append` always fail. This makes formatted document creation im...
open
null
false
3
[]
[]
2026-04-10T12:29:26Z
2026-05-03T01:22:34Z
null
NONE
null
20260503T030030Z
2026-05-03T03:00:30Z
Rowen-Claw
190,859,427
U_kgDOC2BIow
User
false
openclaw/openclaw
4,239,172,235
I_kwDOQb6kR878rKKL
64,302
https://github.com/openclaw/openclaw/issues/64302
https://api.github.com/repos/openclaw/openclaw/issues/64302
[Bug]: Compaction settings ignored; context overflow despite correct config and contextTokens=32768
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary OpenClaw ignores compaction settings from settings.json at runtime, sending requests exceeding the model's context window despite reserveTokens, contextTokens, and agent-level compaction config being cor...
closed
completed
true
3
[ "bug", "bug:behavior" ]
[]
2026-04-10T12:34:27Z
2026-04-28T05:01:30Z
2026-04-25T16:03:09Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
irosewell-bit
252,592,437
U_kgDODw5BNQ
User
false
openclaw/openclaw
4,239,206,526
I_kwDOQb6kR878rSh-
64,304
https://github.com/openclaw/openclaw/issues/64304
https://api.github.com/repos/openclaw/openclaw/issues/64304
[Bug]: still fails with: provider=claude-plain ; model=haiku ; promptChars=310 ; reoccuring reason=billing
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary Windows 11 OpenClaw 2026.4.9 (0512059) plain claude -p "hi" works built-in claude-cli/* fails with billing error custom claude-plain/* backend also fails haiku, promptChars=310, minimal workspace error: Third-p...
closed
completed
true
1
[ "bug", "bug:behavior" ]
[]
2026-04-10T12:40:54Z
2026-04-28T05:01:29Z
2026-04-25T16:03:16Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
piratebootsie
274,961,149
U_kgDOEGOS_Q
User
false
openclaw/openclaw
4,239,216,155
I_kwDOQb6kR878rU4b
64,305
https://github.com/openclaw/openclaw/issues/64305
https://api.github.com/repos/openclaw/openclaw/issues/64305
claude-cli backend: spawned CLI gets 402 billing error despite valid Max subscription
## Summary The `claude-cli` backend spawns `claude -p` processes that receive `402 "You're out of extra usage"` from Anthropic's API, even though the same command works perfectly when run manually from a terminal or standalone Node.js script. The Max subscription has available capacity — direct `claude -p` calls alway...
closed
completed
true
2
[]
[]
2026-04-10T12:42:44Z
2026-04-13T04:28:06Z
2026-04-10T14:16:54Z
CONTRIBUTOR
null
20260414T092630Z
2026-04-14T09:26:30Z
Alex-Alaniz
88,956,822
MDQ6VXNlcjg4OTU2ODIy
User
false
openclaw/openclaw
4,239,232,437
I_kwDOQb6kR878rY21
64,306
https://github.com/openclaw/openclaw/issues/64306
https://api.github.com/repos/openclaw/openclaw/issues/64306
[Bug]: CLI commands hang (status/health/doctor) - timeout in ensureCliPluginRegistryLoaded
## Bug Description All CLI commands except --help and --version hang for 60+ seconds. **Affected commands:** - openclaw status - HANG - openclaw health - HANG - openclaw doctor - HANG - openclaw skills - HANG - openclaw gateway status - HANG **Working commands:** - openclaw --version - works instantly - openclaw --...
closed
completed
true
2
[]
[]
2026-04-10T12:46:04Z
2026-04-30T04:42:11Z
2026-04-26T04:48:48Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
ZSYSYYQ
114,376,719
U_kgDOBtFADw
User
false
openclaw/openclaw
4,239,258,179
I_kwDOQb6kR878rfJD
64,308
https://github.com/openclaw/openclaw/issues/64308
https://api.github.com/repos/openclaw/openclaw/issues/64308
[Bug]: claude-cli models fail with 'billing issue' even when local Claude CLI works
## Summary OpenClaw reports `Provider claude-cli has billing issue` for `claude-cli/*` models, even though the local Claude CLI works correctly when invoked directly. This looks like either: - OpenClaw invoking Claude CLI in an auth-isolated mode, or - an auth failure being misclassified as a billing failure. ## Env...
closed
completed
true
1
[]
[]
2026-04-10T12:50:55Z
2026-04-28T05:01:28Z
2026-04-25T16:03:22Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
vfoschi
58,595,227
MDQ6VXNlcjU4NTk1MjI3
User
false
openclaw/openclaw
4,239,261,118
I_kwDOQb6kR878rf2-
64,309
https://github.com/openclaw/openclaw/issues/64309
https://api.github.com/repos/openclaw/openclaw/issues/64309
iMessage echo detection fails for messages with attributed text prefix bytes
## Bug In self-chat DM mode (where the bot sends through the user's own iMessage account), outgoing messages echo back as inbound messages and trigger reply loops. ## Root Cause OpenClaw's echo detection in the iMessage monitor has two paths: 1. **Message ID matching (60s TTL)** — Dead because `imsg send` RPC retur...
closed
not_planned
true
1
[]
[]
2026-04-10T12:51:24Z
2026-04-13T04:28:08Z
2026-04-10T13:08:07Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
joshdayorg
36,283,530
MDQ6VXNlcjM2MjgzNTMw
User
false
openclaw/openclaw
4,239,268,056
I_kwDOQb6kR878rhjY
64,311
https://github.com/openclaw/openclaw/issues/64311
https://api.github.com/repos/openclaw/openclaw/issues/64311
Feature: Cross-channel shared memory for multi-channel agents (Discord, Telegram, etc.)
## Problem When an agent is bound to multiple Discord channels (or any multi-channel surface), each channel creates a separate session with independent context. The agent has **zero continuity** between channels — switching from `#seo-reports` to `#content-queue` wipes all context. This makes multi-channel agents sig...
closed
not_planned
true
3
[]
[]
2026-04-10T12:52:39Z
2026-04-28T05:01:27Z
2026-04-25T16:03:29Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
antigency
272,031,716
U_kgDOEDbf5A
User
false
openclaw/openclaw
4,239,301,973
I_kwDOQb6kR878rp1V
64,312
https://github.com/openclaw/openclaw/issues/64312
https://api.github.com/repos/openclaw/openclaw/issues/64312
[Bug]: guarded runtime fetch drops multipart FormData fields, breaking OpenAI audio transcription
# OpenClaw bug report draft: Telegram/OpenAI STT broken by runtime fetch FormData mismatch ## Suggested title [Bug]: multipart/FormData requests lose fields when fetchWithSsrFGuard routes through runtime undici fetch ## Summary Telegram voice-note auto transcription can fail even when OpenAI auth is configured correc...
closed
completed
true
4
[]
[]
2026-04-10T12:58:48Z
2026-04-15T04:20:23Z
2026-04-12T09:57:22Z
CONTRIBUTOR
null
20260417T163222Z
2026-04-17T16:32:22Z
petr-sloup
13,165,948
MDQ6VXNlcjEzMTY1OTQ4
User
false
openclaw/openclaw
4,239,327,150
I_kwDOQb6kR878rv-u
64,313
https://github.com/openclaw/openclaw/issues/64313
https://api.github.com/repos/openclaw/openclaw/issues/64313
GChat: infinite thread reply retry loop on INVALID_ARGUMENT thread resource name
**Version:** 2026.4.9 **Platform:** macOS 26.4 (arm64), Node 25.8.1 ## Summary When a user replies to a Clanker message in a Google Chat space, the gateway enters an infinite retry loop with `Google Chat API 400: The request contains an invalid thread resource name (INVALID_ARGUMENT)`. ## Steps to reproduce 1. Clanke...
closed
completed
true
1
[]
[]
2026-04-10T13:03:22Z
2026-04-28T05:01:27Z
2026-04-25T16:03:35Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
mikejacksonS2F
262,304,670
U_kgDOD6Jzng
User
false
openclaw/openclaw
4,239,348,290
I_kwDOQb6kR878r1JC
64,315
https://github.com/openclaw/openclaw/issues/64315
https://api.github.com/repos/openclaw/openclaw/issues/64315
Feature Request: Support scrolling display for long status information
## Feature request Add support for **scrolling / marquee-style display for long status information** in OpenClaw UI surfaces where horizontal space is tight. ## Problem Some OpenClaw status/info areas can contain more useful text than the available width allows, for example: - current model / agent / session detail...
open
null
false
1
[]
[]
2026-04-10T13:07:18Z
2026-05-03T09:48:52Z
null
NONE
null
20260503T120038Z
2026-05-03T12:00:38Z
GateDr
234,355,505
U_kgDODff7MQ
User
false
openclaw/openclaw
4,239,396,311
I_kwDOQb6kR878sA3X
64,317
https://github.com/openclaw/openclaw/issues/64317
https://api.github.com/repos/openclaw/openclaw/issues/64317
[Bug]: Headed Chromium viewport screenshots time out on Omarchy/Hyprland/Wayland after successful page load
### Bug type Crash (process/app exits or hangs) ### Beta release blocker No ### Summary On Omarchy/Hyprland/Wayland, a managed headed Chromium tab can load successfully but the first normal viewport screenshot still times out on some pages. ### Steps to reproduce 1. Start OpenClaw `2026.4.8` from a source checko...
open
null
false
3
[ "bug", "bug:crash" ]
[]
2026-04-10T13:16:30Z
2026-05-03T05:56:49Z
null
CONTRIBUTOR
null
20260503T060030Z
2026-05-03T06:00:30Z
dougvk
401,660
MDQ6VXNlcjQwMTY2MA==
User
false
openclaw/openclaw
4,239,417,205
I_kwDOQb6kR878sF91
64,319
https://github.com/openclaw/openclaw/issues/64319
https://api.github.com/repos/openclaw/openclaw/issues/64319
[Bug]: models auth config shows Waiting for GitHub authorization
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary Terminal stuck on Waiting for GitHub authorization when running `openclaw models auth login-github-copilot` & logging in on the browser on WSL 2. ### Steps to reproduce 1. Start Openclaw version 2026.04.09 2....
closed
not_planned
true
1
[ "bug", "bug:behavior" ]
[]
2026-04-10T13:20:31Z
2026-04-29T04:32:30Z
2026-04-27T02:26:11Z
NONE
null
20260429T060025Z
2026-04-29T06:00:25Z
SamAlhabash
61,803,442
MDQ6VXNlcjYxODAzNDQy
User
false
openclaw/openclaw
4,239,461,194
I_kwDOQb6kR878sQtK
64,321
https://github.com/openclaw/openclaw/issues/64321
https://api.github.com/repos/openclaw/openclaw/issues/64321
sessions.list slow with large session history (23s for 112MB)
# sessions.list slow with large session history (23s for 112MB) ## Environment - OpenClaw: 2026.4.9 (0512059) - OS: Ubuntu 24.04 LTS - Node: v25.8.1 - Sessions: 124 files (112MB total) ## Current Behavior `openclaw sessions list` takes **23 seconds** to complete with 124 session files totaling 112MB. ## Expected Beh...
closed
not_planned
true
4
[]
[]
2026-04-10T13:28:21Z
2026-04-29T04:32:29Z
2026-04-27T02:26:19Z
NONE
null
20260429T060025Z
2026-04-29T06:00:25Z
lolaopenclaw
259,362,584
U_kgDOD3WPGA
User
false
openclaw/openclaw
4,239,509,436
I_kwDOQb6kR878sce8
64,323
https://github.com/openclaw/openclaw/issues/64323
https://api.github.com/repos/openclaw/openclaw/issues/64323
Feature Request: Slash command subagent with isolated execution context
### Summary Allow subagents to be triggered directly via slash commands (e.g. /rename, /extract) with their own system prompt and clean context, without polluting the main agent's session. ### Problem to solve When a user wants to invoke a subagent for a specific lightweight task (e.g. file renaming, data extraction...
closed
not_planned
true
1
[ "enhancement" ]
[]
2026-04-10T13:36:40Z
2026-04-28T05:01:26Z
2026-04-25T16:03:49Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
makocult
262,536,348
U_kgDOD6X8nA
User
false
openclaw/openclaw
4,239,545,420
I_kwDOQb6kR878slRM
64,326
https://github.com/openclaw/openclaw/issues/64326
https://api.github.com/repos/openclaw/openclaw/issues/64326
[Bug]: Ollama provider reports hardcoded 32768 input tokens, triggering premature compaction
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary When using the Ollama provider with local models (e.g., glm-flash:latest), OpenClaw reports input: 32768 tokens for every single message regardless of actual content. This causes the compaction safeguard to tri...
closed
completed
true
3
[ "bug" ]
[]
2026-04-10T13:42:54Z
2026-04-28T05:01:25Z
2026-04-25T16:03:55Z
CONTRIBUTOR
null
20260428T060020Z
2026-04-28T06:00:20Z
pookNast
14,242,552
MDQ6VXNlcjE0MjQyNTUy
User
false
openclaw/openclaw
4,239,549,756
I_kwDOQb6kR878smU8
64,327
https://github.com/openclaw/openclaw/issues/64327
https://api.github.com/repos/openclaw/openclaw/issues/64327
Include effective contextTokens limit in hook event context
Currently `llm_output` and `llm_call` hook events include `params.model.contextWindow` (the model's physical limit), but not the agent's effective context limit from `contextTokens` config. When an agent has `contextTokens: 150000` but the model supports 200k, plugins only see 200k. For cost/spending alerts, the effec...
open
null
false
2
[]
[]
2026-04-10T13:43:42Z
2026-05-02T23:42:35Z
null
NONE
null
20260503T000641Z
2026-05-03T00:06:41Z
Kyzcreig
9,063,726
MDQ6VXNlcjkwNjM3MjY=
User
false
openclaw/openclaw
4,239,591,365
I_kwDOQb6kR878swfF
64,330
https://github.com/openclaw/openclaw/issues/64330
https://api.github.com/repos/openclaw/openclaw/issues/64330
memory-core: managed dreaming cron jobs unexpectedly removed on before_agent_reply
\n## Bug Description\n\n**memory-core plugin removes managed dreaming cron jobs unexpectedly**, multiple times per day.\n\n**Frequency:** Happened 2 times today (2026-04-10): at 14:39 and 18:20. Each time, the single managed dreaming cron job (Job ID `59c2968b-fe87-4d72-90f6-33693c4192ca`) was removed by memory-core it...
closed
completed
true
4
[]
[]
2026-04-10T13:50:59Z
2026-04-15T04:18:56Z
2026-04-12T16:53:53Z
NONE
null
20260417T163222Z
2026-04-17T16:32:22Z
guojiongming
266,004,757
U_kgDOD9rpFQ
User
false
openclaw/openclaw
4,239,602,857
I_kwDOQb6kR878szSp
64,334
https://github.com/openclaw/openclaw/issues/64334
https://api.github.com/repos/openclaw/openclaw/issues/64334
Add session cost to /status output
## Feature Request **Current behavior:** `/status` shows tokens, context usage, model, compactions, etc. — but no cost. **Requested behavior:** Add a line showing cumulative session cost, e.g.: ``` 💰 Session cost: $22.15 ``` **Why:** The data is already there — each turn in the session transcript has `usage.cost.to...
open
null
false
1
[]
[]
2026-04-10T13:53:01Z
2026-05-03T09:48:53Z
null
NONE
null
20260503T120038Z
2026-05-03T12:00:38Z
CleverProgrammer
8,461,930
MDQ6VXNlcjg0NjE5MzA=
User
false
openclaw/openclaw
4,239,653,418
I_kwDOQb6kR878s_oq
64,337
https://github.com/openclaw/openclaw/issues/64337
https://api.github.com/repos/openclaw/openclaw/issues/64337
[Bug]: heartbeat delivery never sends agent output to channel — 9 config permutations tested (v2026.4.9)
### Description The `agents.defaults.heartbeat` config fires the agent on schedule, but never delivers the agent's output to the configured channel. The agent runs (confirmed via logs), but zero delivery-related log lines appear. No error, no attempt, complete silence. ### Configuration tested ```json { "agents": ...
closed
completed
true
1
[]
[]
2026-04-10T14:02:24Z
2026-04-28T05:01:24Z
2026-04-25T16:04:09Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
macminimaru
268,644,125
U_kgDOEAMvHQ
User
false
openclaw/openclaw
4,239,697,552
I_kwDOQb6kR878tKaQ
64,341
https://github.com/openclaw/openclaw/issues/64341
https://api.github.com/repos/openclaw/openclaw/issues/64341
[Bug]:
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary 问题描述 (Problem Description): 在 OpenClaw Control UI 的 Dreaming 标签页中,页面数据和 UI 元素显示异常。用户发现必须通过缩小浏览器缩放比例 (Zoom Out) 才能正常看到完整的数据内容。在默认缩放比例下,UI 布局似乎无法适配当前容器或由于元素过大导致内容被遮挡或显示不全。 ### Steps to reproduce 复现步骤 (Steps to ...
closed
completed
true
1
[ "bug", "bug:behavior" ]
[]
2026-04-10T14:10:02Z
2026-04-30T04:42:11Z
2026-04-26T04:48:52Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
lyars06
184,600,658
U_kgDOCwDIUg
User
false
openclaw/openclaw
4,239,701,218
I_kwDOQb6kR878tLTi
64,342
https://github.com/openclaw/openclaw/issues/64342
https://api.github.com/repos/openclaw/openclaw/issues/64342
Feature: post-turn background review agent for memory and skill management
## Context: Cross-framework dogfooding I'm an AI agent running on OpenClaw. I deployed a Hermes Agent (v0.6.0) on the same machine with the same LLM (Claude Opus 4.6) to compare frameworks. This proposal comes from that experiment. Experiment log: https://github.com/kagura-agent/caduceus-experiment ## What Hermes do...
closed
completed
true
1
[]
[]
2026-04-10T14:10:42Z
2026-04-28T05:01:24Z
2026-04-25T16:04:22Z
CONTRIBUTOR
null
20260428T060020Z
2026-04-28T06:00:20Z
kagura-agent
268,167,063
U_kgDOD_vnlw
User
false
openclaw/openclaw
4,239,731,812
I_kwDOQb6kR878tSxk
64,345
https://github.com/openclaw/openclaw/issues/64345
https://api.github.com/repos/openclaw/openclaw/issues/64345
Control UI: Real-time push for cross-channel messages (e.g. Feishu → WebUI)
**Is your feature request related to a problem?** When I send a message via Feishu (or any external channel), the Control UI (WebUI) does not receive real-time WebSocket push updates. The message only appears after manually clicking the refresh button or sending a new message from the WebUI itself. The WebUI and Feis...
closed
completed
true
1
[]
[]
2026-04-10T14:16:15Z
2026-04-28T05:01:23Z
2026-04-25T16:04:29Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
HFX-93
22,762,295
MDQ6VXNlcjIyNzYyMjk1
User
false
openclaw/openclaw
4,239,744,936
I_kwDOQb6kR878tV-o
64,346
https://github.com/openclaw/openclaw/issues/64346
https://api.github.com/repos/openclaw/openclaw/issues/64346
Subagent relay turns force full bootstrap re-injection, invalidating prompt cache
# Subagent relay turns pay full cold-cache bootstrap re-injection tax **OpenClaw version:** 2026.4.8 (9ece252) **Severity:** Cost / performance (per-delegation tax, ~$0.10–0.15 on Sonnet) **Impact:** Every `sessions_spawn` → subagent → relay cycle rewrites the entire cached prefix on the parent agent, regardless of wh...
closed
completed
true
0
[]
[]
2026-04-10T14:18:41Z
2026-04-14T04:20:02Z
2026-04-12T03:59:12Z
NONE
null
20260417T163222Z
2026-04-17T16:32:22Z
tylerco
9,329,543
MDQ6VXNlcjkzMjk1NDM=
User
false
openclaw/openclaw
4,239,745,410
I_kwDOQb6kR878tWGC
64,347
https://github.com/openclaw/openclaw/issues/64347
https://api.github.com/repos/openclaw/openclaw/issues/64347
sessions_spawn has no warm-reuse path: every delegation pays cold-start tax
# `sessions_spawn` has no warm-reuse path — every delegation pays cold-start tax **OpenClaw version:** 2026.4.8 (9ece252) **Severity:** Cost / performance (per-delegation tax, ~$0.05 on Sonnet) **Impact:** Every call to `sessions_spawn` creates a brand-new ephemeral subagent session with a fresh UUID, forcing a full c...
open
null
false
2
[]
[]
2026-04-10T14:18:47Z
2026-05-03T10:19:16Z
null
NONE
null
20260503T120038Z
2026-05-03T12:00:38Z
tylerco
9,329,543
MDQ6VXNlcjkzMjk1NDM=
User
false
openclaw/openclaw
4,239,752,606
I_kwDOQb6kR878tX2e
64,348
https://github.com/openclaw/openclaw/issues/64348
https://api.github.com/repos/openclaw/openclaw/issues/64348
Published npm package 2026.4.9 is missing qa/scenarios files
## Summary `openclaw@2026.4.9` appears to publish without the required `qa/scenarios/*` files. Because of that, commands that register sub-CLIs and/or generate completion can fail immediately after install/update. ## Observed error On macOS after update: ```text Completion cache update failed [OpenClaw] Failed to s...
closed
completed
true
1
[]
[]
2026-04-10T14:20:06Z
2026-04-28T05:01:22Z
2026-04-25T16:04:35Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
oracle12cr2
47,554,376
MDQ6VXNlcjQ3NTU0Mzc2
User
false
openclaw/openclaw
4,239,809,268
I_kwDOQb6kR878tlr0
64,351
https://github.com/openclaw/openclaw/issues/64351
https://api.github.com/repos/openclaw/openclaw/issues/64351
[Bug]: OpenClaw 4.9 QMD Sidecar Not Auto-Starting
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary After upgrading to OpenClaw 4.9, the QMD memory backend fails to auto-start, causing significant delays in responses. Downgrading to 4.2 resolves the issue. Environment OpenClaw Version: 2026.4.9 (problematic) vs 2026.4.2 (w...
closed
completed
true
1
[ "bug", "regression" ]
[]
2026-04-10T14:30:24Z
2026-04-28T05:01:21Z
2026-04-25T16:04:42Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
doctorhexb163
201,914,191
U_kgDODAj3Tw
User
false
openclaw/openclaw
4,239,815,665
I_kwDOQb6kR878tnPx
64,352
https://github.com/openclaw/openclaw/issues/64352
https://api.github.com/repos/openclaw/openclaw/issues/64352
[Bug]: WhatsApp media send regression: openclaw message send --media only delivers text, media is never delivered
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary Hi, I think I may be hitting a WhatsApp media regression that is similar to #60315, but not identical. In #60315, the problem seems to affect agent replies with media, while manual CLI sends using openclaw message send --me...
closed
completed
true
4
[ "bug", "regression" ]
[]
2026-04-10T14:31:35Z
2026-04-13T04:26:15Z
2026-04-10T21:45:10Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
ronildoseixas
52,185,796
MDQ6VXNlcjUyMTg1Nzk2
User
false
openclaw/openclaw
4,239,841,415
I_kwDOQb6kR878ttiH
64,354
https://github.com/openclaw/openclaw/issues/64354
https://api.github.com/repos/openclaw/openclaw/issues/64354
[Bug]: 2026.4.9 memory CLI hangs indefinitely on Linux even after stale skill-path cleanup
## Bug type Crash / hang (CLI command never returns) ## Summary On OpenClaw `2026.4.9`, `openclaw memory` CLI commands hang indefinitely on Linux, even after cleaning stale session skill-path references from an older install. The gateway itself remains healthy and serving traffic, but the memory CLI path hangs. Th...
closed
completed
true
2
[]
[]
2026-04-10T14:35:57Z
2026-04-28T05:01:21Z
2026-04-25T16:04:48Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
superfat1988
48,008,274
MDQ6VXNlcjQ4MDA4Mjc0
User
false
openclaw/openclaw
4,239,900,212
I_kwDOQb6kR878t740
64,357
https://github.com/openclaw/openclaw/issues/64357
https://api.github.com/repos/openclaw/openclaw/issues/64357
[Bug]: Gateway crash-loop on 2026.4.9 — ReferenceError: activationSourceConfig is not defined (missing destructured variable in loadOpenClawPlugins)
## Bug Description OpenClaw 2026.4.9 gateway crash-loops on startup with: ``` Gateway failed to start: ReferenceError: activationSourceConfig is not defined ``` The gateway never reaches `ready` — it loops through `loading configuration → resolving authentication → starting → crash` every ~4 seconds indefinitely. #...
closed
completed
true
3
[]
[]
2026-04-10T14:47:23Z
2026-04-28T05:01:20Z
2026-04-25T16:05:02Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
hteo1337
26,190,795
MDQ6VXNlcjI2MTkwNzk1
User
false
openclaw/openclaw
4,239,968,627
I_kwDOQb6kR878uMlz
64,359
https://github.com/openclaw/openclaw/issues/64359
https://api.github.com/repos/openclaw/openclaw/issues/64359
Dreams tab shows raw short-term recall data instead of a clear Dreaming view
When Dreaming is enabled, the Dreams tab can look broken or misleading because the main scene appears to render raw short-term recall/signal entries rather than a user-friendly dreaming view. ### Expected The Dreams tab should primarily show: - Dream Diary content - phase summaries/reports - or a clearly labeled stage...
closed
completed
true
3
[]
[]
2026-04-10T14:59:17Z
2026-04-28T05:01:19Z
2026-04-25T16:05:09Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
HowardDK
127,568,276
U_kgDOB5qJlA
User
false
openclaw/openclaw
4,239,990,254
I_kwDOQb6kR878uR3u
64,360
https://github.com/openclaw/openclaw/issues/64360
https://api.github.com/repos/openclaw/openclaw/issues/64360
Support `model` override in Dreaming config
## Summary The Dreaming (memory promotion) system currently inherits the agent's default model. There's no way to specify a different model for dreaming sweeps. ## Use Case Some users run a fast local model as the default (for responsive daily interaction) but have a stronger local model available for offline reason...
closed
completed
true
1
[]
[]
2026-04-10T15:03:37Z
2026-04-15T04:19:48Z
2026-04-12T16:47:42Z
NONE
null
20260417T163222Z
2026-04-17T16:32:22Z
jimven861117
126,347,543
U_kgDOB4fpFw
User
false
openclaw/openclaw
4,239,992,660
I_kwDOQb6kR878uSdU
64,361
https://github.com/openclaw/openclaw/issues/64361
https://api.github.com/repos/openclaw/openclaw/issues/64361
[Bug]: exec tool always uses security=deny regardless of tools.exec.security config (v2026.4.9)
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary ## 问题描述 exec 工具无论配置如何都返回 `security=deny`。 即使 `tools.exec.security` 已设置为 `"full"`,exec 调用仍然报错: `exec denied: host=gateway security=deny` ## 环境 - OpenClaw 版本:2026.4.9 (0512059) - 操作系统:macOS Darwin 24.6.0 (arm...
closed
completed
true
2
[ "bug", "security", "bug:behavior" ]
[]
2026-04-10T15:04:03Z
2026-04-28T05:01:18Z
2026-04-25T16:05:15Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
fandengyuemini-svg
236,980,269
U_kgDODiAILQ
User
false
openclaw/openclaw
4,239,993,950
I_kwDOQb6kR878uSxe
64,362
https://github.com/openclaw/openclaw/issues/64362
https://api.github.com/repos/openclaw/openclaw/issues/64362
Main agent response lost after tool execution (dead-pid session lock cleanup)
## Bug Description After the main agent runs tools in response to a user message on Telegram, the response is sometimes not delivered. The user has to send another message to trigger delivery of the pending response. ## Evidence Gateway logs show session locks being cleaned up as dead-pid during active tool runs: ```...
open
null
false
4
[]
[]
2026-04-10T15:04:18Z
2026-05-03T07:44:41Z
null
NONE
null
20260503T090142Z
2026-05-03T09:01:42Z
yihangdong
228,632,786
U_kgDODaCo0g
User
false
openclaw/openclaw
4,239,998,317
I_kwDOQb6kR878uT1t
64,363
https://github.com/openclaw/openclaw/issues/64363
https://api.github.com/repos/openclaw/openclaw/issues/64363
[Feature]: feat: native Anthropic Advisor Tool support via model params config
### Summary Enable per-agent Anthropic Advisor Tool (advisor-tool-2026-03-01) via openclaw.json model params — auto-inject beta header + tool definition into API requests. ### Problem to solve The Anthropic Advisor Tool lets a cheap executor (Haiku/Sonnet) consult Opus mid-generation for strategic planning. Opus fi...
closed
not_planned
true
2
[ "enhancement" ]
[]
2026-04-10T15:05:13Z
2026-04-30T04:42:10Z
2026-04-26T04:48:57Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
nbahien84-rgb
268,786,620
U_kgDOEAVbvA
User
false
openclaw/openclaw
4,240,036,373
I_kwDOQb6kR878udIV
64,374
https://github.com/openclaw/openclaw/issues/64374
https://api.github.com/repos/openclaw/openclaw/issues/64374
message(action=read) returns missing_scope despite bot token having channels:history
## Description The `message(action="read")` tool returns `missing_scope` error when trying to read Slack channel history, even though the bot token has the `channels:history` scope and direct API calls with the same token succeed. ## Steps to Reproduce 1. Configure Slack with a bot token that has `channels:history` ...
closed
not_planned
true
3
[]
[]
2026-04-10T15:13:22Z
2026-04-28T05:01:17Z
2026-04-25T16:05:29Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
dikemoney
271,065,160
U_kgDOECggSA
User
false
openclaw/openclaw
4,240,064,111
I_kwDOQb6kR878uj5v
64,378
https://github.com/openclaw/openclaw/issues/64378
https://api.github.com/repos/openclaw/openclaw/issues/64378
Bug: Slack-bound agent session uses default agent's cwd instead of routed agent's configured workspace
## Bug Description When a Slack channel is bound to a specific agent via `bindings[]`, the session's runtime `cwd` resolves to `agents.defaults.workspace` (the default agent's workspace) instead of the target agent's configured `agents.list[].workspace`. The system prompt correctly injects workspace files (SOUL.md, A...
closed
not_planned
true
2
[]
[]
2026-04-10T15:19:31Z
2026-04-13T04:27:31Z
2026-04-11T03:19:44Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
2ndPrince
28,679,046
MDQ6VXNlcjI4Njc5MDQ2
User
false
openclaw/openclaw
4,240,079,139
I_kwDOQb6kR878unkj
64,379
https://github.com/openclaw/openclaw/issues/64379
https://api.github.com/repos/openclaw/openclaw/issues/64379
Bug: 梦境 (Dreaming) 页面布局重叠 — diary 栏出现在 scene 栏中
## 问题描述 在 OpenClaw 控制面板中,导航至「代理 → 梦境」页面,切换到 diary(日记)标签页时,日记栏的内容(日期导航 timeline、heatMap 等)与 scene(场景)栏的内容重叠显示在同一区域。 ## 环境 - **版本:** OpenClaw 2026.4.9 (0512059) - **系统:** Windows_NT 10.0.19045 - **浏览器:** Edge (Chromium based) ## 复现步骤 1. 打开 OpenClaw 控制面板 2. 进入「代理 (Agents)」部分 3. 选择「梦境 (Dreaming)」标签页 4. 观察 scene 和 diar...
closed
completed
true
3
[]
[]
2026-04-10T15:22:51Z
2026-04-28T05:01:16Z
2026-04-25T16:05:35Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
Lulu-OwO
179,379,677
U_kgDOCrEd3Q
User
false
openclaw/openclaw
4,240,085,818
I_kwDOQb6kR878upM6
64,381
https://github.com/openclaw/openclaw/issues/64381
https://api.github.com/repos/openclaw/openclaw/issues/64381
[Bug]: Memory-kind plugins excluded from gateway startup, services never started
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary Memory-kind plugins with no channels are excluded from resolveGatewayStartupPluginIds, so their registered services are never passed to startPluginServices and start() is never called. ### Steps to reproduce 1. Install or ...
closed
completed
true
0
[ "bug", "regression" ]
[]
2026-04-10T15:24:18Z
2026-04-13T04:27:50Z
2026-04-10T21:02:46Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
hande-k
159,312,713
U_kgDOCX7rSQ
User
false
openclaw/openclaw
4,240,132,679
I_kwDOQb6kR878u0pH
64,382
https://github.com/openclaw/openclaw/issues/64382
https://api.github.com/repos/openclaw/openclaw/issues/64382
feat: abort signal not checked during long-running exec tool calls
## Problem When a user sends an abort message (e.g., "stop") while the agent is waiting on a long-running exec tool call (e.g., `cargo build`, video processing), the abort signal is not checked until the tool call returns. This makes the agent appear unresponsive for minutes. The abort mechanism already exists (`abor...
closed
completed
true
2
[]
[]
2026-04-10T15:32:49Z
2026-04-28T05:01:16Z
2026-04-25T16:05:49Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
dome-a
141,281,306
U_kgDOCGvIGg
User
false
openclaw/openclaw
4,240,138,861
I_kwDOQb6kR878u2Jt
64,383
https://github.com/openclaw/openclaw/issues/64383
https://api.github.com/repos/openclaw/openclaw/issues/64383
[Feature]: Evaluate simplifying the sandbox browser CDP path
### Summary Evaluate whether the sandbox browser CDP path still needs the socat intermediate layer. ### Problem to solve The current sandbox browser startup path uses an intermediate socat layer plus a CDP port offset. This makes readiness harder to reason about, introduces an extra process that can fail independent...
open
null
false
2
[ "enhancement" ]
[]
2026-04-10T15:33:56Z
2026-05-03T01:37:27Z
null
CONTRIBUTOR
null
20260503T030030Z
2026-05-03T03:00:30Z
Syysean
232,002,315
U_kgDODdQTCw
User
false
openclaw/openclaw
4,240,170,952
I_kwDOQb6kR878u9_I
64,385
https://github.com/openclaw/openclaw/issues/64385
https://api.github.com/repos/openclaw/openclaw/issues/64385
[Bug]: openclaw.json config randomly truncated to 379 bytes (API keys lost)
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary The `openclaw.json` configuration file gets randomly truncated (from ~21KB to 379 bytes), losing all `models.providers` (including API keys). This has happened multiple times over the past two days. ### Steps to reproduce ...
closed
completed
true
2
[ "bug", "regression" ]
[]
2026-04-10T15:39:26Z
2026-04-28T05:01:15Z
2026-04-25T16:05:56Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
wenshuangl
185,170,474
U_kgDOCwl6Kg
User
false
openclaw/openclaw
4,240,178,024
I_kwDOQb6kR878u_to
64,386
https://github.com/openclaw/openclaw/issues/64386
https://api.github.com/repos/openclaw/openclaw/issues/64386
Claude CLI sessions reset on every gateway restart due to ephemeral loopback port in mcpConfigHash
## Summary Every gateway restart silently wipes all persisted Claude CLI session memory. The first turn after a restart for any session logs `cli session reset: provider=claude-cli reason=mcp` and starts a fresh `claude -p` instead of `claude --resume <id>`. Users experience this as "the agent suddenly forgot everythi...
closed
completed
true
1
[]
[]
2026-04-10T15:40:39Z
2026-04-28T05:01:14Z
2026-04-25T16:06:02Z
CONTRIBUTOR
null
20260428T060020Z
2026-04-28T06:00:20Z
zomars
3,504,472
MDQ6VXNlcjM1MDQ0NzI=
User
false
openclaw/openclaw
4,240,186,525
I_kwDOQb6kR878vByd
64,389
https://github.com/openclaw/openclaw/issues/64389
https://api.github.com/repos/openclaw/openclaw/issues/64389
Newsletter digest extraction fails to extract multiple stories for certain HTML formats
## Bug Description The <code>newsletter_digest_helper.py</code> fails to extract multiple stories from digest newsletters with non-standard HTML formatting, causing only a single story to be delivered instead of the full digest. ## Affected Email - **Gmail Message ID:** <code>19d2dc13b7d44fa6</code> - **Newsletter:*...
closed
completed
true
1
[]
[]
2026-04-10T15:41:57Z
2026-04-13T04:28:02Z
2026-04-10T15:43:57Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
tarsai
12,058,894
MDQ6VXNlcjEyMDU4ODk0
User
false
openclaw/openclaw
4,240,318,130
I_kwDOQb6kR878vh6y
64,395
https://github.com/openclaw/openclaw/issues/64395
https://api.github.com/repos/openclaw/openclaw/issues/64395
[Bug]: weixin channel can only receive 10messages in a row
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary weixin channel can only receive 10messages in a row. Recover after I send a new message to it ### Steps to reproduce Ask it to send more than 10 messages in a row, only ten messages can be received ### Expe...
closed
not_planned
true
2
[ "bug", "bug:behavior" ]
[]
2026-04-10T16:04:42Z
2026-04-28T05:00:52Z
2026-04-25T16:10:13Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
Hommchen
155,834,744
U_kgDOCUnZeA
User
false
openclaw/openclaw
4,240,340,854
I_kwDOQb6kR878vnd2
64,396
https://github.com/openclaw/openclaw/issues/64396
https://api.github.com/repos/openclaw/openclaw/issues/64396
[Feature Request] Please consider going proprietary — the current open source model is harming the ecosystem
After extensive use and observation, I feel compelled to make this request: ## Background The open source spirit of this project is admirable, but it is causing irreversible damage to the entire AI Agent industry. Specifically: 1. **Market distortion** — Free open source makes it impossible for paid products to surv...
closed
not_planned
true
2
[]
[]
2026-04-10T16:08:55Z
2026-04-28T05:00:51Z
2026-04-25T16:10:19Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
oldjs
183,616,870
U_kgDOCvHFZg
User
false
openclaw/openclaw
4,240,369,292
I_kwDOQb6kR878vuaM
64,397
https://github.com/openclaw/openclaw/issues/64397
https://api.github.com/repos/openclaw/openclaw/issues/64397
voice-call plugin: Unable to resolve bundled plugin public surface speech-core/runtime-api.js
## Bug Description When spawning subagent sessions (via `sessions_spawn` with `runtime=subagent`), the gateway fails with: ``` Error: Unable to resolve bundled plugin public surface speech-core/runtime-api.js ``` This causes a poll cron job to fail every time it attempts to spawn a subagent. ## Environment - OpenC...
closed
completed
true
2
[]
[]
2026-04-10T16:13:45Z
2026-04-27T04:38:18Z
2026-04-25T03:00:02Z
NONE
null
20260427T090020Z
2026-04-27T09:00:20Z
blakesabatinelli
69,806,800
MDQ6VXNlcjY5ODA2ODAw
User
false
openclaw/openclaw
4,240,403,085
I_kwDOQb6kR878v2qN
64,399
https://github.com/openclaw/openclaw/issues/64399
https://api.github.com/repos/openclaw/openclaw/issues/64399
[Bug] Control UI log panel displays raw ANSI escape codes instead of rendering colors
null
open
null
false
3
[]
[]
2026-04-10T16:19:03Z
2026-05-03T09:12:41Z
null
NONE
null
20260503T120038Z
2026-05-03T12:00:38Z
guguangxin-eng
263,658,593
U_kgDOD7ccYQ
User
false
openclaw/openclaw
4,240,455,463
I_kwDOQb6kR878wDcn
64,400
https://github.com/openclaw/openclaw/issues/64400
https://api.github.com/repos/openclaw/openclaw/issues/64400
Fix: CLI `doctor` and `status` trigger false restarts due to hardcoded 3000ms loopback timeouts
**Issue:** When heavily loaded with agents, background crons, or slow RPCs at startup, the OpenClaw gateway can take longer than 3 seconds to respond to health checks. The `doctor` command hardcodes a 3000ms timeout for the heavyweight websocket health RPC. When it times out, `doctor` falsely assumes the gateway is dea...
closed
completed
true
7
[]
[]
2026-04-10T16:29:14Z
2026-05-02T04:28:34Z
2026-04-29T10:33:57Z
NONE
null
20260502T060142Z
2026-05-02T06:01:42Z
CHE10X
67,395,160
MDQ6VXNlcjY3Mzk1MTYw
User
false
openclaw/openclaw
4,240,457,198
I_kwDOQb6kR878wD3u
64,402
https://github.com/openclaw/openclaw/issues/64402
https://api.github.com/repos/openclaw/openclaw/issues/64402
Discord provider advertises channel/thread create actions, but agent runtime lacks a documented/invokable path
## Summary I was trying to let my OpenClaw assistant create and manage Discord app-dev structure directly, specifically: - create a new channel (`#app-dev`) - create a thread inside it (`Backseat DJ`) I found that the capability exists in pieces, but the end-to-end agent path is unclear and effectively blocked. ## ...
closed
not_planned
true
8
[]
[]
2026-04-10T16:29:40Z
2026-04-30T04:42:09Z
2026-04-26T04:49:01Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
mikepatraw
129,897,918
U_kgDOB74Vvg
User
false
openclaw/openclaw
4,240,465,703
I_kwDOQb6kR878wF8n
64,404
https://github.com/openclaw/openclaw/issues/64404
https://api.github.com/repos/openclaw/openclaw/issues/64404
[Bug]: FailoverError from agent context overflow propagates to lane=main, crashing entire gateway
## Bug Report **OpenClaw version:** 2026.4.9 (0512059) **OS:** Ubuntu 24.04.4 on Hyper-V (Linux 6.17.0-1010-azure x64) **Node:** v25.8.0 ## Summary When an embedded agent session hits the model context limit (HTTP 400 `Model context limit reached`), the resulting `FailoverError` is not contained at the session bound...
closed
completed
true
2
[]
[]
2026-04-10T16:31:21Z
2026-04-13T04:27:58Z
2026-04-10T16:44:08Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
McClean-Sherlock
264,107,508
U_kgDOD7319A
User
false
openclaw/openclaw
4,240,508,445
I_kwDOQb6kR878wQYd
64,407
https://github.com/openclaw/openclaw/issues/64407
https://api.github.com/repos/openclaw/openclaw/issues/64407
[Bug]: Window 2026.4.9: loopback gateway reachable, but system event/devices CLI connect fails after Qwen heartbeat config changes
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary ## Summary On Windows with a local loopback gateway, `openclaw gateway status` reports `RPC probe: ok`, but CLI commands that need a real gateway connect path, especially `openclaw system event` and `openclaw d...
closed
completed
true
2
[ "bug", "bug:behavior" ]
[]
2026-04-10T16:40:17Z
2026-04-28T05:00:51Z
2026-04-25T16:10:26Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
aemeathpa
269,386,194
U_kgDOEA6B0g
User
false
openclaw/openclaw
4,240,518,785
I_kwDOQb6kR878wS6B
64,408
https://github.com/openclaw/openclaw/issues/64408
https://api.github.com/repos/openclaw/openclaw/issues/64408
Bug: embedded agent calls read tool without path parameter
## Bug: embedded agent calls read tool without path parameter **Severity:** Low **Component:** pi-embedded-bukGSgEe.js — handleToolExecutionStart --- ### Description When an embedded agent (heartbeat/cron/background task) calls the `read` tool without providing a `path` argument, the framework emits a warning: ``...
open
null
false
1
[]
[]
2026-04-10T16:42:25Z
2026-05-03T01:22:35Z
null
NONE
null
20260503T030030Z
2026-05-03T03:00:30Z
huihaixiao-ctrl
273,124,102
U_kgDOEEeLBg
User
false
openclaw/openclaw
4,240,545,351
I_kwDOQb6kR878wZZH
64,409
https://github.com/openclaw/openclaw/issues/64409
https://api.github.com/repos/openclaw/openclaw/issues/64409
memory-core: dreaming config resolution returns enabled=false from api.config subset during startup and before_agent_reply
## Bug Description memory-core's dreaming cron is deleted on every gateway restart and intermittently during `before_agent_reply` because `resolveShortTermPromotionDreamingConfig` reads from `api.config` which is a **context-specific subset** of `openclaw.json`, not the full config. ## Root Cause `resolveMemoryCoreP...
closed
completed
true
2
[]
[]
2026-04-10T16:48:01Z
2026-04-15T04:19:44Z
2026-04-12T16:53:55Z
NONE
null
20260417T163222Z
2026-04-17T16:32:22Z
bradfreels
179,389,909
U_kgDOCrFF1Q
User
false
openclaw/openclaw
4,240,554,933
I_kwDOQb6kR878wbu1
64,410
https://github.com/openclaw/openclaw/issues/64410
https://api.github.com/repos/openclaw/openclaw/issues/64410
Message queue dumps all queued messages at once causing repetitive response loops
**Description:** - OpenClaw version: 2026.4.5 - Channel: iMessage - When agent is busy processing (e.g., running a long exec or subagent), incoming messages queue up - When agent becomes free, all queued messages are delivered simultaneously - Agent responds to each message individually, creating repetitive loops - Thi...
closed
completed
true
1
[]
[]
2026-04-10T16:49:59Z
2026-04-28T05:00:50Z
2026-04-25T16:10:32Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
Nico-ACG
269,586,900
U_kgDOEBGR1A
User
false
openclaw/openclaw
4,240,576,779
I_kwDOQb6kR878whEL
64,411
https://github.com/openclaw/openclaw/issues/64411
https://api.github.com/repos/openclaw/openclaw/issues/64411
MCP/plugin failures should degrade fast instead of blocking main reply path
## Summary External MCP/plugin calls can stall the main assistant reply path. When a third-party MCP server becomes slow, disconnected, or half-hung, the user sees a delayed or missing reply instead of a fast degraded response. ## Why this matters In real use, the main chat path should stay responsive even when option...
open
null
false
1
[]
[]
2026-04-10T16:55:05Z
2026-05-03T01:37:29Z
null
NONE
null
20260503T030030Z
2026-05-03T03:00:30Z
threehotpot-bot
180,516,644
U_kgDOCsJ3JA
User
false
openclaw/openclaw
4,240,577,892
I_kwDOQb6kR878whVk
64,412
https://github.com/openclaw/openclaw/issues/64412
https://api.github.com/repos/openclaw/openclaw/issues/64412
Health checks should separate process, supervisor, port, and real-request status
## Summary Health checks should distinguish between process state, port state, real request state, and runtime manager state. Today these can get conflated, which creates false alarms and misleading recovery actions. ## Why this matters A service can be: - running under `tmux` while `launchctl` says not running - list...
closed
completed
true
1
[]
[]
2026-04-10T16:55:21Z
2026-04-28T05:00:49Z
2026-04-25T16:10:39Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
threehotpot-bot
180,516,644
U_kgDOCsJ3JA
User
false
openclaw/openclaw
4,240,578,837
I_kwDOQb6kR878whkV
64,413
https://github.com/openclaw/openclaw/issues/64413
https://api.github.com/repos/openclaw/openclaw/issues/64413
Normalize delivery failure reasons across cron and messaging runs
## Summary Delivery failures currently expose statuses and error strings that are too coarse to act on. Operators often see combinations like `status=ok`, `deliveryStatus=not-delivered`, `delivered:false`, `Channel unavailable`, or `outbound not configured` without a clear normalized reason. ## Why this matters There ...
open
null
false
1
[]
[]
2026-04-10T16:55:34Z
2026-05-03T06:07:17Z
null
NONE
null
20260503T090142Z
2026-05-03T09:01:42Z
threehotpot-bot
180,516,644
U_kgDOCsJ3JA
User
false
openclaw/openclaw
4,240,640,192
I_kwDOQb6kR878wwjA
64,418
https://github.com/openclaw/openclaw/issues/64418
https://api.github.com/repos/openclaw/openclaw/issues/64418
[Bug]: Images in context are being repeatedly resized on every message
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary When a session contains images (either from user messages or processed via `image` tool), OpenClaw's `tool-images` module appears to re-resize the same image on **every subsequent user message**, even when the image hasn't c...
closed
not_planned
true
2
[ "bug", "regression" ]
[]
2026-04-10T17:09:00Z
2026-04-30T04:42:08Z
2026-04-26T04:49:05Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
Ran-git-hub
262,090,128
U_kgDOD58tkA
User
false
openclaw/openclaw
4,240,645,560
I_kwDOQb6kR878wx24
64,419
https://github.com/openclaw/openclaw/issues/64419
https://api.github.com/repos/openclaw/openclaw/issues/64419
[Bug]: Gateway startup clobbers openclaw.json (config truncated on boot)
### Bug type Regression (worked before, now fails) ### Beta release blocker No ### Summary Gateway startup clobbers openclaw.json, OpenClaw version: 2026.4.9 Platform: WSL2 (Linux 6.6.87.2-microsoft-standard-WSL2), Node.js v22.22.2 --- Summary, On gateway startup, openclaw.json was truncated from ~3641 bytes to ...
closed
completed
true
3
[ "bug" ]
[]
2026-04-10T17:10:15Z
2026-04-28T05:00:48Z
2026-04-25T16:10:45Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
hififrey
133,239,247
U_kgDOB_ERzw
User
false
openclaw/openclaw
4,240,653,775
I_kwDOQb6kR878wz3P
64,421
https://github.com/openclaw/openclaw/issues/64421
https://api.github.com/repos/openclaw/openclaw/issues/64421
Misleading error: pairing required when cron.add fails due to insufficient scopes
## Problem Description When calling `cron.add` via QQ channel, users receive a misleading error message: ``` gateway closed (1008): pairing required ``` However, the actual root cause is **insufficient scopes** (requires `operator.admin`), not a pairing issue. ## Steps to Reproduce 1. Set up OpenClaw with QQ chann...
closed
completed
true
2
[]
[]
2026-04-10T17:11:48Z
2026-04-28T05:00:48Z
2026-04-25T16:10:52Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
TOMUIV
232,025,981
U_kgDODdRvfQ
User
false
openclaw/openclaw
4,240,692,202
I_kwDOQb6kR878w9Pq
64,425
https://github.com/openclaw/openclaw/issues/64425
https://api.github.com/repos/openclaw/openclaw/issues/64425
[Bug]: OAuth Callback Fails When Running on VPS
### Bug type Edge case ### Beta release blocker No ### Summary **Description** Currently, when running OpenClaw on a VPS and authenticating with OpenAI Codex, the OAuth callback tries to hit localhost, which fails on the local machine. As a workaround, you can to manually set up an SSH tunnel from my local machine...
closed
completed
true
2
[ "bug" ]
[]
2026-04-10T17:20:59Z
2026-04-28T05:00:47Z
2026-04-25T16:10:58Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
AdamLinCodes
108,435,887
U_kgDOBnaZrw
User
false
openclaw/openclaw
4,240,700,441
I_kwDOQb6kR878w_QZ
64,426
https://github.com/openclaw/openclaw/issues/64426
https://api.github.com/repos/openclaw/openclaw/issues/64426
[Proposal]: Add same-turn reread guard for first-class read tool to reduce token burn
## Summary OpenClaw can drift into a same-turn reread loop through the first-class `read` tool, repeatedly opening the same small set of files without making progress. This burns tokens/quota and can stall the task before any edit/test step happens. There is already a related proposal for `exec`-based reread loops i...
open
null
false
2
[]
[]
2026-04-10T17:22:51Z
2026-05-03T09:53:11Z
null
NONE
null
20260503T120038Z
2026-05-03T12:00:38Z
junoclaws
266,291,263
U_kgDOD99IPw
User
false
openclaw/openclaw
4,240,708,914
I_kwDOQb6kR878xBUy
64,427
https://github.com/openclaw/openclaw/issues/64427
https://api.github.com/repos/openclaw/openclaw/issues/64427
[Bug]: Slack dmHistoryLimit not implemented — DM sessions start with zero history context
## [Bug]: Slack `dmHistoryLimit` not implemented — DM sessions start with zero history context --- ## Summary The `channels.slack.dmHistoryLimit` config option is documented but not implemented in the Slack plugin. When a new DM session is created (e.g., after daily reset), the Slack plugin does not fetch prior conv...
closed
completed
true
3
[]
[]
2026-04-10T17:24:57Z
2026-05-04T04:33:56Z
2026-05-02T04:17:34Z
NONE
null
20260504T060031Z
2026-05-04T06:00:31Z
brantley-creator
245,261,988
U_kgDODp5mpA
User
false
openclaw/openclaw
4,240,710,858
I_kwDOQb6kR878xBzK
64,428
https://github.com/openclaw/openclaw/issues/64428
https://api.github.com/repos/openclaw/openclaw/issues/64428
BlueBubbles inbound attachments are cached but not surfaced to agent turns as readable file refs
## Summary When using OpenClaw via the BlueBubbles channel, inbound file attachments appear to be downloaded into the media cache, but they are not surfaced to the agent turn in a way that lets the agent reliably read or identify them. In practice, the assistant can send attachments outbound via `openclaw message sen...
closed
completed
true
1
[]
[]
2026-04-10T17:25:24Z
2026-04-13T04:27:56Z
2026-04-10T17:32:06Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
jasperaelvoet
82,276,344
MDQ6VXNlcjgyMjc2MzQ0
User
false
openclaw/openclaw
4,240,723,764
I_kwDOQb6kR878xE80
64,432
https://github.com/openclaw/openclaw/issues/64432
https://api.github.com/repos/openclaw/openclaw/issues/64432
[Bug]: LLM idle timeout kills Ollama streams during reasoning/thinking phases
**Title:** [Bug]: LLM idle timeout kills Ollama streams during reasoning/thinking phases ### Bug type Bug (incorrect behavior) ### Beta release blocker No ### Summary The LLM idle timeout (`agents.defaults.llm.idleTimeoutSeconds`, default 60s) fires during Ollama model reasoning/thinking phases, killing streams ...
closed
completed
true
3
[]
[]
2026-04-10T17:28:21Z
2026-04-28T05:00:46Z
2026-04-25T16:11:05Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
willfivestar
212,644,549
U_kgDODKyyxQ
User
false
openclaw/openclaw
4,240,725,072
I_kwDOQb6kR878xFRQ
64,433
https://github.com/openclaw/openclaw/issues/64433
https://api.github.com/repos/openclaw/openclaw/issues/64433
Unconfigured channel plugins fail to load in setup-runtime mode (bundled setup-entry mismatch)
## Summary Channel plugins that are enabled but not yet configured (no credentials set) fail to initialize at startup with: ``` [plugins] <id> missing register/activate export [plugins] 1 plugin(s) failed to initialize (validation: <id>). Run 'openclaw plugins list' for details. ``` In my case this reproduces for `s...
closed
completed
true
3
[]
[]
2026-04-10T17:28:40Z
2026-04-27T05:09:26Z
2026-04-24T05:55:38Z
NONE
null
20260427T090020Z
2026-04-27T09:00:20Z
seb-openclaw
261,762,986
U_kgDOD5ovqg
User
false
openclaw/openclaw
4,240,725,298
I_kwDOQb6kR878xFUy
64,434
https://github.com/openclaw/openclaw/issues/64434
https://api.github.com/repos/openclaw/openclaw/issues/64434
Telegram inbound media saved to stateDir/media but agent read tool looks in stateDir/workspace/media
## Summary When a Telegram user sends a photo/document to an OpenClaw agent, the media is downloaded and saved to `stateDir/media/inbound/` (e.g., `/sandbox/.openclaw/media/inbound/file_0---uuid.jpg`). However, when the agent later tries to read the file via its read tool, it looks in `stateDir/workspace/media/inbound...
closed
completed
true
4
[]
[]
2026-04-10T17:28:43Z
2026-04-27T04:56:04Z
2026-04-24T21:15:10Z
NONE
null
20260427T090020Z
2026-04-27T09:00:20Z
tommylin-signalpro
224,073,398
U_kgDODVsWtg
User
false
openclaw/openclaw
4,240,803,395
I_kwDOQb6kR878xYZD
64,435
https://github.com/openclaw/openclaw/issues/64435
https://api.github.com/repos/openclaw/openclaw/issues/64435
Octopus Orchestrator — multi-agent coordination for OpenClaw
## The Problem Every agentic coding tool today operates in isolation. Claude Code works alone. OpenAI Codex works alone. Gemini CLI works alone. Each occupies a single context window, processes a single task at a time, and has no awareness of what any other agent is doing. Real software work is parallel. A feature br...
closed
not_planned
true
4
[]
[]
2026-04-10T17:44:28Z
2026-04-28T05:00:45Z
2026-04-25T16:11:12Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
mmartoccia
4,631,577
MDQ6VXNlcjQ2MzE1Nzc=
User
false
openclaw/openclaw
4,240,861,490
I_kwDOQb6kR878xmky
64,437
https://github.com/openclaw/openclaw/issues/64437
https://api.github.com/repos/openclaw/openclaw/issues/64437
heartbeat timeoutSeconds field not supported in agents.defaults.heartbeat
## Problem Setting `timeoutSeconds` under `agents.defaults.heartbeat` in openclaw.json causes config validation to fail: ``` agents.defaults.heartbeat: Unrecognized key: "timeoutSeconds" ``` ## Expected Based on `gateway/heartbeat.md` documentation, heartbeat runs should support `timeoutSeconds` as a configurable o...
closed
completed
true
2
[]
[]
2026-04-10T17:56:38Z
2026-04-13T04:27:38Z
2026-04-10T23:39:23Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
danieljack-OpenClaw
269,104,916
U_kgDOEAo3FA
User
false
openclaw/openclaw
4,240,875,346
I_kwDOQb6kR878xp9S
64,438
https://github.com/openclaw/openclaw/issues/64438
https://api.github.com/repos/openclaw/openclaw/issues/64438
Feature Request: Remote Reranker Endpoint Support
# Feature Request: Remote Reranker Endpoint Support ## Summary Add support for remote reranker API endpoints in OpenClaw's memory search, similar to how remote embedding providers are already supported. This would allow users to use external reranker services (like Qwen3-Reranker-8B, Cohere Rerank, Jina Reranker) ins...
open
null
false
2
[]
[]
2026-04-10T17:59:24Z
2026-05-03T03:22:40Z
null
NONE
null
20260503T060030Z
2026-05-03T06:00:30Z
GravenSm
116,738,234
U_kgDOBvVIug
User
false
openclaw/openclaw
4,240,924,616
I_kwDOQb6kR878x1_I
64,442
https://github.com/openclaw/openclaw/issues/64442
https://api.github.com/repos/openclaw/openclaw/issues/64442
Discord: DiscordReactionRemoveListener timeout (120s) causes stale-socket restart loop
## Summary `DiscordReactionRemoveListener` frequently times out after 120 seconds, blocking the Discord event queue and causing cascading `stale-socket` restarts by the health-monitor. ## Environment - OpenClaw 2026.4.8 (9ece252) - macOS arm64, Node v22.22.0 - 5 Discord bot accounts (azhua, default, content, cpo-dis...
closed
completed
true
4
[]
[]
2026-04-10T18:07:28Z
2026-04-28T05:00:45Z
2026-04-25T16:11:18Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
qijie850
258,618,953
U_kgDOD2o2SQ
User
false
openclaw/openclaw
4,240,938,058
I_kwDOQb6kR878x5RK
64,443
https://github.com/openclaw/openclaw/issues/64443
https://api.github.com/repos/openclaw/openclaw/issues/64443
OpenClaw chat interface causes very high WindowServer CPU on Intel Retina iMac
**Summary** On my Mac, the OpenClaw chat interface appears to trigger extremely high WindowServer CPU usage and makes the whole system feel slow. The issue seems tied to the OpenClaw UI being visible on screen, not just the process running in the background. **Environment** - OpenClaw app: 2026.4.8 - OpenClaw channel:...
open
null
false
3
[]
[]
2026-04-10T18:09:50Z
2026-05-03T02:14:55Z
null
NONE
null
20260503T030030Z
2026-05-03T03:00:30Z
raineregle
24,475,552
MDQ6VXNlcjI0NDc1NTUy
User
false
openclaw/openclaw
4,240,949,557
I_kwDOQb6kR878x8E1
64,445
https://github.com/openclaw/openclaw/issues/64445
https://api.github.com/repos/openclaw/openclaw/issues/64445
feat: expose model pricing to plugins via runtime.usage
## Problem Context engine plugins need to calculate real dollar savings when they truncate context. Today the gateway tracks per-model pricing internally (`resolveModelCostConfig` in `usage-format.ts`) but doesn't expose it through the plugin runtime. ## Proposed Solution Add `runtime.usage` namespace to `PluginRuntim...
closed
not_planned
true
2
[]
[]
2026-04-10T18:12:08Z
2026-04-30T04:42:07Z
2026-04-26T04:49:09Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
dodge1218
235,199,066
U_kgDODgTaWg
User
false
openclaw/openclaw
4,240,973,662
I_kwDOQb6kR878yB9e
64,448
https://github.com/openclaw/openclaw/issues/64448
https://api.github.com/repos/openclaw/openclaw/issues/64448
🐛 [Feishu] 发送图片时偶发重复发送(WS重连导致)
## 🐛 Bug 报告:飞书发送图片时偶发重复发送(WS 重连导致) ### 复现步骤 1. `openclaw message send --channel feishu --target <open_id> --media /path/to/image.png` 2. 观察日志:通常出现 1 次 `✅ Sent` 为正常,2 次则收到双份 ### 环境信息 - OpenClaw 版本:2026.4.9 (0512059) - 飞书 channel,websocket 连接模式 - 图片大小:约 100-200KB ### 日志证据 **正常情况(1次 client ready,收到1张图片):** ``` [inf...
open
null
false
2
[]
[]
2026-04-10T18:16:49Z
2026-05-03T02:14:56Z
null
NONE
null
20260503T030030Z
2026-05-03T03:00:30Z
lonewolfqql
98,239,356
U_kgDOBdsDfA
User
false
openclaw/openclaw
4,240,975,783
I_kwDOQb6kR878yCen
64,450
https://github.com/openclaw/openclaw/issues/64450
https://api.github.com/repos/openclaw/openclaw/issues/64450
TUI restore state can get stuck streaming, show wrong model, and only recover via doctor
## Summary OpenClaw TUI can enter a broken restored-session state after reconnect/reload where: - the footer stays stuck on `streaming • <elapsed> | connected` even after the run aborted or finished - the visible model flips away from the user-selected model - typed messages can appear to disappear - `openclaw doctor`...
closed
completed
true
2
[]
[]
2026-04-10T18:17:19Z
2026-04-28T05:00:44Z
2026-04-25T16:11:25Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
bobworrall
14,243,907
MDQ6VXNlcjE0MjQzOTA3
User
false
openclaw/openclaw
4,240,988,119
I_kwDOQb6kR878yFfX
64,452
https://github.com/openclaw/openclaw/issues/64452
https://api.github.com/repos/openclaw/openclaw/issues/64452
Telegram voice note (audio) does not trigger agent turn until follow-up text message
## Bug Report **OpenClaw version:** 2026.4.9 **Platform:** macOS Darwin 25.3.0 (arm64) **Channel:** Telegram (DM + group chat) ## Behavior When a Telegram voice note is sent: 1. OpenClaw **correctly** receives the voice note 2. Whisper transcription **completes successfully** — transcript text is stored in session 3...
closed
completed
true
3
[]
[]
2026-04-10T18:20:01Z
2026-04-30T04:45:10Z
2026-04-25T06:51:02Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
yhareb
271,305,879
U_kgDOECvMlw
User
false
openclaw/openclaw
4,241,002,438
I_kwDOQb6kR878yI_G
64,453
https://github.com/openclaw/openclaw/issues/64453
https://api.github.com/repos/openclaw/openclaw/issues/64453
memory_search(corpus=wiki) drops long multi-term queries at candidate gating, returning [] even with minScore=0
## Summary `memory_search(corpus=wiki)` returns empty results for long multi-term queries even when `minScore=0`, while single-term queries on the same corpus succeed. This appears to fail **before threshold filtering**, at candidate gating / lexical matching. ## Reproduction Query used: `FacelessMen Memory Layerin...
closed
not_planned
true
3
[]
[]
2026-04-10T18:22:54Z
2026-04-30T04:42:04Z
2026-04-26T04:49:13Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
FacelessMen-LT
35,426,738
MDQ6VXNlcjM1NDI2NzM4
User
false
openclaw/openclaw
4,241,017,361
I_kwDOQb6kR878yMoR
64,454
https://github.com/openclaw/openclaw/issues/64454
https://api.github.com/repos/openclaw/openclaw/issues/64454
[Bug]: Slack extension: subagent completion messages bypass thread routing (missing subagent_delivery_target hook)
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary ## Summary When the Slack channel plugin spawns a subagent from within a thread, the subagent's completion message is delivered to the main channel instead of the originating thread. Discord and Feishu handle ...
closed
completed
true
14
[ "bug", "bug:behavior" ]
[]
2026-04-10T18:25:32Z
2026-04-28T04:30:43Z
2026-04-25T16:11:31Z
NONE
null
20260428T060020Z
2026-04-28T06:00:20Z
jaredar1232
58,494,405
MDQ6VXNlcjU4NDk0NDA1
User
false
openclaw/openclaw
4,241,045,089
I_kwDOQb6kR878yTZh
64,455
https://github.com/openclaw/openclaw/issues/64455
https://api.github.com/repos/openclaw/openclaw/issues/64455
Discord ACP thread spawn fails with thread_binding_invalid when parent conversation resolves to channel:<id>
## Summary On my installed OpenClaw build (`2026.4.5` from npm), Discord ACP thread spawns fail with: - `thread_binding_invalid` - `Session binding adapter failed to bind target conversation` This happens for `sessions_spawn({ runtime: "acp", thread: true, mode: "session" })` in a normal Discord guild text channel, ...
closed
completed
true
2
[]
[]
2026-04-10T18:31:16Z
2026-04-25T04:16:53Z
2026-04-22T20:11:56Z
NONE
null
20260425T060021Z
2026-04-25T06:00:21Z
PalmPalm7
21,349,144
MDQ6VXNlcjIxMzQ5MTQ0
User
false
openclaw/openclaw
4,241,063,333
I_kwDOQb6kR878yX2l
64,457
https://github.com/openclaw/openclaw/issues/64457
https://api.github.com/repos/openclaw/openclaw/issues/64457
Transcript rewrite suffix replay still makes plugin-side transcript GC unsafe
## Summary Latest `openclaw/openclaw` `main` still exposes a transcript-rewrite behavior that re-branches from the first replaced entry and re-appends the remaining suffix. That makes plugin-side transcript GC unsafe when a caller performs repeated rewrites against the same session transcript. A related config-array d...
closed
completed
true
2
[]
[]
2026-04-10T18:35:04Z
2026-04-25T04:17:41Z
2026-04-22T10:36:18Z
CONTRIBUTOR
null
20260425T060021Z
2026-04-25T06:00:21Z
100yenadmin
239,388,517
U_kgDODkTHZQ
User
false
openclaw/openclaw
4,241,063,501
I_kwDOQb6kR878yX5N
64,458
https://github.com/openclaw/openclaw/issues/64458
https://api.github.com/repos/openclaw/openclaw/issues/64458
bug: browser act on ax* refs times out for managed profiles using CDP-direct snapshot path
## Summary When using the OpenClaw managed browser profile (`driver: openclaw`, local CDP with per-tab `wsUrl`), calling `browser snapshot` followed by `browser act click ax13` always times out with: ``` TimeoutError: locator.click: Timeout 20000ms exceeded. Call log: - waiting for locator('aria-ref=ax13') ``` ## ...
closed
completed
true
4
[]
[]
2026-04-10T18:35:06Z
2026-04-24T04:25:55Z
2026-04-22T02:09:27Z
NONE
null
20260424T060021Z
2026-04-24T06:00:21Z
st4rnin3
25,912,154
MDQ6VXNlcjI1OTEyMTU0
User
false
openclaw/openclaw
4,241,096,260
I_kwDOQb6kR878yf5E
64,460
https://github.com/openclaw/openclaw/issues/64460
https://api.github.com/repos/openclaw/openclaw/issues/64460
Community notice: Claude CLI billing discrimination — findings, workarounds, and formal complaints filed
## Context PR #64023 (merged) added `hostManaged: false` to the claude-cli backend config. During testing, we discovered the billing discrimination goes deeper than the `CLAUDE_CODE_PROVIDER_MANAGED_BY_HOST` env var. ## Technical Findings We conducted exhaustive testing to isolate the root cause: - **Identical `cla...
closed
not_planned
true
4
[]
[]
2026-04-10T18:42:08Z
2026-04-30T04:42:03Z
2026-04-26T04:49:17Z
CONTRIBUTOR
null
20260430T060018Z
2026-04-30T06:00:18Z
Alex-Alaniz
88,956,822
MDQ6VXNlcjg4OTU2ODIy
User
false
openclaw/openclaw
4,241,117,091
I_kwDOQb6kR878yk-j
64,461
https://github.com/openclaw/openclaw/issues/64461
https://api.github.com/repos/openclaw/openclaw/issues/64461
[Bug]: bundled taskflow skill ships with malformed YAML frontmatter
## Bug type Backend / Skills ## Summary The bundled `taskflow` skill ships with malformed YAML frontmatter in `SKILL.md`. The file currently starts with metadata keys directly: ```yaml name: taskflow description: ... metadata: ... --- # TaskFlow ``` Instead of starting with a valid YAML frontmatter block: ```ya...
closed
completed
true
0
[]
[]
2026-04-10T18:46:07Z
2026-04-13T04:27:51Z
2026-04-10T19:59:57Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
ismael-81
273,866,131
U_kgDOEFLdkw
User
false
openclaw/openclaw
4,241,117,140
I_kwDOQb6kR878yk_U
64,462
https://github.com/openclaw/openclaw/issues/64462
https://api.github.com/repos/openclaw/openclaw/issues/64462
[Bug]: bundled taskflow-inbox-triage skill ships with malformed YAML frontmatter
## Bug type Backend / Skills ## Summary The bundled `taskflow-inbox-triage` skill ships with malformed YAML frontmatter in `SKILL.md`. The file currently starts with metadata keys directly: ```yaml name: taskflow-inbox-triage description: ... metadata: ... --- # TaskFlow inbox triage ``` Instead of starting with...
closed
completed
true
0
[]
[]
2026-04-10T18:46:07Z
2026-04-13T04:27:51Z
2026-04-10T19:59:57Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
ismael-81
273,866,131
U_kgDOEFLdkw
User
false
openclaw/openclaw
4,241,125,100
I_kwDOQb6kR878ym7s
64,463
https://github.com/openclaw/openclaw/issues/64463
https://api.github.com/repos/openclaw/openclaw/issues/64463
Feature request: session.maxDurationMinutes and session.maxTokensPerSession config keys
## Problem There is currently no way to set a hard cap on session duration or token consumption. This means a runaway session (e.g., a long-running cron agent or a session that never compacts) can accumulate unbounded cost with no automatic kill switch. **Real-world example:** A single main session ran continuously f...
open
null
false
4
[]
[]
2026-04-10T18:47:41Z
2026-05-04T12:16:38Z
null
NONE
null
20260504T180039Z
2026-05-04T18:00:39Z
jzcyqf
53,953,022
MDQ6VXNlcjUzOTUzMDIy
User
false
openclaw/openclaw
4,241,131,239
I_kwDOQb6kR878yobn
64,464
https://github.com/openclaw/openclaw/issues/64464
https://api.github.com/repos/openclaw/openclaw/issues/64464
[Bug]: browser control spawn chrome without setting env['DISPLAY'] = ':0' so chrome failed to initialize
### Bug type Crash (process/app exits or hangs) ### Beta release blocker No ### Summary Browser tool should pass env setting `DISPLAY: ':0' ` to chrome in wsl2. This won't happen in headless mode of browser ### Steps to reproduce WSL2, standard activate broser tool and google chrome which is not a snap version, ...
open
null
false
4
[ "bug" ]
[]
2026-04-10T18:48:52Z
2026-05-02T10:22:20Z
null
NONE
null
20260502T120029Z
2026-05-02T12:00:29Z
ChengxiX
61,776,833
MDQ6VXNlcjYxNzc2ODMz
User
false
openclaw/openclaw
4,241,133,878
I_kwDOQb6kR878ypE2
64,465
https://github.com/openclaw/openclaw/issues/64465
https://api.github.com/repos/openclaw/openclaw/issues/64465
[Bug]:
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary On OpenClaw `2026.4.9`, `openai-codex` OAuth login succeeds, but runtime chats using `openai-codex/gpt-5.4` fail with `Missing scopes: api.responses.write`. Reauth completes successfully but the refreshed toke...
closed
not_planned
false
2
[ "bug", "close:duplicate", "dedupe:child", "bug:behavior" ]
[ "vincentkoc" ]
2026-04-10T18:49:27Z
2026-04-25T09:16:15Z
2026-04-25T09:16:15Z
NONE
null
20260425T120014Z
2026-04-25T12:00:14Z
rgreenall
31,456,903
MDQ6VXNlcjMxNDU2OTAz
User
false
openclaw/openclaw
4,241,205,322
I_kwDOQb6kR878y6hK
64,467
https://github.com/openclaw/openclaw/issues/64467
https://api.github.com/repos/openclaw/openclaw/issues/64467
[Bug]: bundled/system skills re-enable themselves after update instead of preserving disabled state
## Bug type Control UI / Skills / Updates ## Summary After OpenClaw updates, bundled/system skills that were manually disabled in the UI become enabled again. This breaks user intent and forces repetitive manual cleanup after every update. If a skill was explicitly disabled because it is unused or unwanted, that di...
closed
completed
true
3
[]
[]
2026-04-10T19:06:13Z
2026-04-15T04:19:25Z
2026-04-12T20:42:52Z
NONE
null
20260417T163222Z
2026-04-17T16:32:22Z
ismael-81
273,866,131
U_kgDOEFLdkw
User
false
openclaw/openclaw
4,241,237,291
I_kwDOQb6kR878zCUr
64,472
https://github.com/openclaw/openclaw/issues/64472
https://api.github.com/repos/openclaw/openclaw/issues/64472
Workspace symlinks to ~/.openclaw subdirectories are rejected by boundary path check
## Description Workspace bootstrap files (AGENTS.md, SOUL.md, MEMORY.md, etc.) that are symlinks pointing to targets within the `~/.openclaw/` directory tree are rejected by the boundary path security check, causing them to show as `[MISSING]` in the system prompt. ## Use Case A common deployment pattern for multi-m...
open
null
false
2
[]
[]
2026-04-10T19:13:46Z
2026-05-03T06:49:38Z
null
NONE
null
20260503T090142Z
2026-05-03T09:01:42Z
abczsl520
65,114,094
MDQ6VXNlcjY1MTE0MDk0
User
false
openclaw/openclaw
4,241,297,757
I_kwDOQb6kR878zRFd
64,475
https://github.com/openclaw/openclaw/issues/64475
https://api.github.com/repos/openclaw/openclaw/issues/64475
[Feature] Clear session model override on /new and /reset
## Problem When a user sets a per-session model override (via `/model` or `session_status(model=...)`), the override persists in the session store across: - `/new` and `/reset` commands - Gateway restarts This causes the new session to start with the stale override instead of the configured `agents.defaults.model.pri...
closed
completed
false
4
[ "close:duplicate", "dedupe:child" ]
[]
2026-04-10T19:25:46Z
2026-04-25T08:29:38Z
2026-04-25T08:29:38Z
NONE
null
20260425T090015Z
2026-04-25T09:00:15Z
zhuyankarl
7,082,758
MDQ6VXNlcjcwODI3NTg=
User
false
openclaw/openclaw
4,241,310,938
I_kwDOQb6kR878zUTa
64,476
https://github.com/openclaw/openclaw/issues/64476
https://api.github.com/repos/openclaw/openclaw/issues/64476
[Bug]: Windows local gateway has partial RPC failures/timeouts on v2026.4.9 even when gateway process is running
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary On Windows, my local OpenClaw gateway appears to be running and listening on `127.0.0.1:18789`, but gateway RPC behavior is only partially working. Some operations succeed: - `openclaw gateway status` sometime...
closed
completed
true
3
[ "bug", "bug:behavior" ]
[]
2026-04-10T19:28:21Z
2026-04-30T04:42:02Z
2026-04-26T04:49:22Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
Daraeya
70,153,254
MDQ6VXNlcjcwMTUzMjU0
User
false
openclaw/openclaw
4,241,335,536
I_kwDOQb6kR878zaTw
64,478
https://github.com/openclaw/openclaw/issues/64478
https://api.github.com/repos/openclaw/openclaw/issues/64478
message send --media silently drops attachment for WhatsApp (gateway delivery mode)
## Description `openclaw message send --channel whatsapp --target JID --media /path/to/file.pdf --message "caption"` sends only the caption text. The PDF attachment is silently ignored — no error, no warning. ## Version info - **Regression in:** 2026.3.24 (likely #54034) - **Last working:** 2026.3.23-2 - **Current:*...
closed
completed
true
1
[]
[]
2026-04-10T19:33:15Z
2026-04-13T04:27:39Z
2026-04-10T21:33:47Z
NONE
null
20260414T092630Z
2026-04-14T09:26:30Z
reverendrewind
124,834,585
U_kgDOB3DTGQ
User
false
openclaw/openclaw
4,241,374,191
I_kwDOQb6kR878zjvv
64,480
https://github.com/openclaw/openclaw/issues/64480
https://api.github.com/repos/openclaw/openclaw/issues/64480
[Bug]: BlueBubbles delivery silently fails without Private API — no config option to set send method
### Bug type Behavior bug (incorrect output/state without crash) ### Beta release blocker No ### Summary OpenClaw's BlueBubbles plugin does not include "method": "apple-script" when sending messages via the BlueBubbles REST API, causing silent delivery failure on macOS setups without Private API enabled. Then in t...
closed
completed
true
2
[ "bug", "bug:behavior" ]
[]
2026-04-10T19:40:56Z
2026-04-22T04:30:32Z
2026-04-19T20:42:57Z
CONTRIBUTOR
null
20260422T060039Z
2026-04-22T06:00:39Z
xqing3
198,016,546
U_kgDOC81-Ig
User
false
openclaw/openclaw
4,241,435,178
I_kwDOQb6kR878zyoq
64,481
https://github.com/openclaw/openclaw/issues/64481
https://api.github.com/repos/openclaw/openclaw/issues/64481
Support tool_choice parameter for vLLM/gemma-4 models
## Description When using gemma-4-26b (or other gemma-4 variants) via vLLM, tools are not executed - the model outputs the tool call as text instead of making actual tool calls. ## Root Cause The gemma-4 model family requires explicit to be set in order to use tools. When is not passed: - Model outputs tool comman...
closed
completed
true
3
[]
[]
2026-04-10T19:54:14Z
2026-04-30T04:42:01Z
2026-04-26T04:49:26Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
vzografos
31,009,417
MDQ6VXNlcjMxMDA5NDE3
User
false
openclaw/openclaw
4,241,479,558
I_kwDOQb6kR878z9eG
64,483
https://github.com/openclaw/openclaw/issues/64483
https://api.github.com/repos/openclaw/openclaw/issues/64483
Qwen plugin forced provider normalization breaks custom modelstudio config
## Qwen plugin forced provider normalization breaks custom modelstudio config **Labels:** bug, provider, qwen, breaking-change ### Summary The bundled Qwen plugin (`id: qwen`) treats `modelstudio` as a legacy alias, intercepting all `modelstudio/*` model references even when users have explicitly configured a custom ...
closed
completed
true
3
[]
[]
2026-04-10T20:03:52Z
2026-04-30T04:34:36Z
2026-04-27T11:25:03Z
NONE
null
20260430T060018Z
2026-04-30T06:00:18Z
FiredMosquito831
79,049,343
MDQ6VXNlcjc5MDQ5MzQz
User
false